Much has been written about the technologies that will drive 5G, particularly how they will improve users’ experience with connectivity. Similarly, much has been said about how ongoing technological developments will usher in a new generation of network-enabled applications.
Much has been written about the technologies that drive 5G, and specifically how those technologies will improve users’ experience with connectivity. Similarly, much has been said about how the technology’s continued development will usher in a new generation of network-enabled applications. In this article, we’ll discuss one key aspect of 5G technology and how it will impact the development of wireless network capacity.
This is one of the more important, yet often overlooked, aspects of the evolution of wireless communications. It represents another key reason why the convergence of cloud computing and wireless communications makes so much sense: Simply put, many of the complex problems associated with 5G wireless networks can be solved using software, eliminating the need for the costly, time-consuming, and often slow-to-evolve hardware used to date.
Cloud and Telecommunications: Ideal for Next-Generation Networks
It is well known that some of the most advanced technologies that make up 5G can be implemented in software running on commodity servers. This is very exciting as it allows for a slow and sure departure from the proprietary hardware that has been used for the last four generations of telecommunications networks. The shift to software will help reduce overall capital and operational costs for telecommunications providers. Equally important, the shift from hardware to software will enable the telecommunications industry to be agile and proactive, future-proofing such networks by rolling out desired features on a regular basis, rather than waiting a decade or so for the next generation of standards to emerge. Innovation will thrive by creating a world where the transition from generation to generation is a software upgrade, just as the cloud industry has done for over a decade.
I will talk more about this in future blogs, but today I want to talk about wireless capacity, or technically, spectral efficiency. I hope you understand that we can use computing power to increase cellular network capacity, and advances in software-based machine learning and data analytics techniques can be used to improve the efficiency of 5G and future networks. When you add this to other elements of the ecosystem, the combination of cloud computing and communications networks is a perfect fit.
5G Core Technology: Massive Multi-User MIMO
Multiple-input multiple-output (MIMO) is a method of using multiple transmit and receive antennas to take advantage of multipath propagation and increase the capacity of a wireless link. MIMO is a key element of Wi-Fi, 3G, and 4G wireless communication standards. However, 5G takes it to the next level with massive multi-user (MU) MIMO, which dramatically increases the number of antennas to support many simultaneous users. This technology is key to 5G’s promise of a 1,000x increase in capacity over 4G.
The science behind Massive MU-MIMO lies in the complex mathematics of manipulating the signals sent and received by all antennas so that the communication channel to each user is maintained and can withstand environmental distortions. This is the subject of many technical books and academic studies, but a simplified version is shown in the diagram below.
Massive MU-MIMO involves a lot of matrix multiplications and transpositions, all of which require huge computations, which are directly related to the number of users a cell tower serves and the number of antennas in the cell tower. Moreover, this calculation is done every few milliseconds for thousands of subcarriers, which means it requires huge amounts of processing power and energy. As network operators increase the number of antennas, the computational requirements increase significantly, along with other related issues.
User patterns also influence the amount of computation required. The precoding method described in the diagram above works best when users are stationary or moving slowly. Otherwise, the precoding matrix needs to be recalculated frequently, requiring many more computations. In this case, an alternative method called “conjugate beamforming” is more effective, but requires the number of antennas to far exceed the number of users, and radio capacity is generally reduced.
Therefore, the overall capacity that a network provides is directly related to how much computing power an operator is willing to purchase and deploy at each of their thousands of cell towers. Edge computing is a great fit for this, as it allows computing to be easily scaled. Even if some operators do not need a lot of capacity right away, it is still important to understand how to architect their network in such a way that it can be easily scaled as demand for network capacity increases.
Microsoft has been investing heavily in computing technologies that can achieve massive MU-MIMO for 5G networks. As early as 2012, Microsoft Research invested in a practical solution to implement MU-MIMO using a distributed pipeline using commodity server racks (edge data centers), meeting timing specifications and allowing it to scale to hundreds of antennas (this technology is state of the art and was reported at SIGCOMM 2013).
Deep Learning for Wireless Capacity
5G is moving towards an open architecture and there are many ways to optimize the network. This approach adds complexity, but deep learning techniques can be used to address these complexities that are usually beyond the capabilities of humans. In the case of precoding for Massive MIMO mentioned above, deep learning techniques can be applied to select algorithms that reduce energy consumption while minimizing capacity loss. With predictive analytics and modern software that adapts to dynamic network loads, 5G networks will become smarter.
Microsoft is investing heavily in machine learning and AI, supporting the research of world-leading experts in the field, and working to enhance communication networks by designing deep learning algorithms with domain knowledge. In addition to the examples above, we are also actively investigating how deep learning techniques can be used to control transmission power to reduce interference and increase capacity.
Continuous machine learning (leveraging flexible edge computing to model dynamic radio frequency environments and user mobility patterns) and management of the signal processing pipeline creates a tremendous value proposition for the telecommunications industry. This major advancement allows research breakthroughs to be rapidly incorporated into the system, not only to expand wireless capacity but also to improve the overall operational efficiency of 5G networks.
Azure: Where edge computing, cloud, and telcos converge
Microsoft has been investing heavily in edge computing for over a decade and continues to do so. In particular, Azure is committed to delivering compute closer to cell towers, where it will be of greatest benefit to network operators looking to extend their networks in a cost-effective way. Additionally, through our Azure for Operators initiative, we continue to enable new first and third-party solutions that further enhance and simplify edge computing, from network connectivity to on-demand computing to full orchestration.
Given Microsoft’s ability to scale up computing in line with carrier demand, the power of edge technologies, including massive MU-MIMO, is exactly what carriers have been looking for. Azure helps carriers achieve their goal of increasing capacity as their networks grow and evolve. While carriers increase the number of antennas and base stations, Microsoft can stand up servers at scale and manage them from anywhere in the world, making Azure a perfect fit for 5G and beyond carrier networks.
learn more
To learn more, read the Azure for Operators e-book.